US could reach 1,000,000 coronavirus cases today, definitely tomorrow.

[Ms. Toad]

14. You're not looking at rate of increase, you're looking at absolute numbers of new cases.

Last edited Mon Apr 27, 2020, 03:25 PM - Edit history (1)

(And you're also picking out a single day (which is dependent on a lot of things, and subject to change for a couple of weeks as they sort out where the new cases really belong due to the changed definitions).

The rate of increase began slowing down around 3/22.

Here is the daily multiplier (divide the total number of cases by the total number the day before). Note the downward trend from teh first day (March 22) to the last (April 26). The increase in cases has been in a downward trend for more than a month, from the number of cases growing at a rate of 36% (or higher) to now growing at a rate of 2%. I've bolded the interruptions in the downward trend.

1.360962567
1.353634578
1.388969521
1.31661442
1.280952381
1.307311029
1.30521327
1.180464779
1.250692095
1.266847024
1.241312367
1.184860807
1.166710665
1.174793529
1.135689522
1.127618079
1.167543992
1.139443514
1.119326211
1.112917887
1.091841365
1.07177292
1.06692776
1.093258223
1.087463848
1.067039618
1.072854892
1.050283059
1.039917622
1.047724994
1.062609386
1.051783204
1.048857895
1.058864514
1.020022184
1.02132483

Looking strictly at the increase in new cases (as opposed to the total number of cases) shows the same thing - but it's a bit more erratic because the numbers are smaller (so more statistical noise). But you can still see a trend. I've bolded the days the number of new cases actually went down, since I think it makes the trend easier to see.

1.935894802
1.098386754
1.083099816
1.205013828
1.293699563
1.07805883
1.038910712
1.031730622
1.148821315
1.085312756
1.070146171
1.128832306
1.077466873
1.053343099
0.746096127
1.22708034
1.072023037
0.954748059
1.049530052
1.004212573
0.889964505
0.915221834
0.971872074
1.010190847
1.119625825
0.9822745
1.081654436
0.8948748
0.899858713
1.077279516
0.927233301
1.156111026
1.057467834
1.221675186
0.909158581
0.748440103

So despite single day highs (like 3 days ago), you can see that the growth of new cases decreasing is decreasing - and even when it is increasing (a numer greater than 1), the rate of increase (numbers to the right of the decimal) is going down.

Further - the number are in flux as states are still adjusting to the new definition. The cases from 3 days ago are likely to be spread backwards at least to some extent. The process is that they all come in on the day reported - BUT - then they are adjusted backward to the diagnosis (most likely the death that was reclassified). The number of new cases for close to a month back were recently adjusted up by about 2000 a day (with a corresponding decrease in the most recent reports).

My numbers include the adjustments - which is why people who have seen the first list of numbers before may be surprised becasue there are more interruptions in the downward trend, reflecting the additionof later-added cases.

Keep in mind, you need to look at trends - not absolute numbers on any day. There are far too many variables for a single day's data to be statistically significant. Look at at least 3 days befor eyou decide the trend is changing.

For example the increase in new cases on March 22 (1.95) looks extremely scary - it nearly doubled in a single day. but if you look at the days surrounding, it puts it in context: 1.234010041, 0.860958783, 1.935894802, 1.098386754. Bouncing around, for sure - but when you compare the March 22 growth to the surrounding days it becomes clear it is a statistical oddity, not a trend.